The volume of opinions given in social media is growing, so we need useful tools to analyze them and use the gained information in the future. Opinion mining is an important research field to analyze the opinion given by someone for an entity. The most used methods in opinion mining are machine learning techniques. Our work aims to evaluate the performance of machine learning techniques through experiments in performing opinion classification tasks in the Al-banian language. Our approach to opinion mining addresses the problem of classifying text document opinions as positive or negative opinions. Since a lack of research on the Albanian language in this field, we conducted an experimental evaluation of fourteen techniques used for opinion mining. We tested different machine learning algorithm's performance using Weka. We have presented other conclusions related to the best feature combination of traditional machine learning algorithms.

Keywords : Opinion Mining, Sentiment Analysis, Machine Learning, Albanian Language